Antiviral immune responses present a major hurdle to the efficacious use of oncolytic adenoviruses as cancer treatments. Despite the existence of a highly immunosuppressive tumor environment, adenovirus-infected cells can nonetheless be efficiently cleared by infiltrating cytotoxic T lymphocytes (CTL) without compromising tumor burden. In this study, we tested the hypothesis that tumor-infiltrating T cells could be more effectively activated and redirected by oncolytic adenoviruses that were armed with bispecific T-cell–engager (BiTE) antibodies. The oncolytic adenovirus ICOVIR-15K was engineered to express an EGFR-targeting BiTE (cBiTE) antibody under the control of the major late promoter, leading to generation of ICOVIR-15K-cBiTE, which retained its oncolytic properties in vitro. cBiTE expression and secretion was detected in supernatants from ICOVIR-15K-cBiTE–infected cells, and the secreted BiTEs bound specifically to both CD3+ and EGFR+ cells. In cell coculture assays, ICOVIR-15K-cBiTE–mediated oncolysis resulted in robust T-cell activation, proliferation, and bystander cell-mediated cytotoxicity. Notably, intratumoral injection of this cBiTE-expressing adenovirus increased the persistence and accumulation of tumor-infiltrating T cells in vivo, compared with the parental virus lacking such effects. Moreover, in two distinct tumor xenograft models, combined delivery of ICOVIR-15K-cBiTE with peripheral blood mononuclear cells or T cells enhanced the antitumor efficacy achieved by the parental counterpart. Overall, our results show how arming oncolytic adenoviruses with BiTE can overcome key limitations in oncolytic virotherapy. Cancer Res; 77(8); 2052–63. ©2017 AACR.
Oncolytic adenoviruses have gained considerable attention as anticancer agents. Their attractiveness relies on a multimodal mechanism to kill cancer cells that include direct oncolysis, bystander effect when armed with therapeutic transgenes, and immunogenic cell death–mediated antitumor immune responses (1). Clinical trials have demonstrated the safety and potential of oncolytic viruses (2, 3) and highlighted the importance of the immune system for their success in cancer patients. Tumors develop numerous mechanisms to evade antitumor immune responses (4), but oncolytic viruses can revert such immune suppression (5). However, virus replication in the tumor also triggers a potent immune response against viral epitopes, which has been shown to be dominant over tumor-associated epitopes (6, 7). Hence, oncolytic virus-infected cells are often cleared from the tumor microenvironment by infiltrating virus-specific cytotoxic T lymphocytes (CTL) without altering the tumor burden (8). Many efforts to improve oncolytic viruses rely on encoding immunostimulatory molecules (e.g., GM-CSF, IL12, or CD40L) in the virus genome (9, 10). However, these strategies do not address either the virus immunodominance or the infiltrating antivirus CTL.
Bispecific T-cell engagers (BiTE) are novel immunotherapeutic molecules composed of two single-chain antibodies (scFV) connected through a flexible peptide linker. One of the scFV is specific for a tumor-associated antigen on target cells, whereas the second scFV is specific for the CD3 T-cell coreceptor. This format allows the transient binding of BiTE molecules to T cells and target cells simultaneously, leading to T-cell activation and specific target cell lysis (11). The potential of BiTEs has been demonstrated with the CD19-targeting BiTE blinatumomab, approved by the FDA for the treatment of acute lymphoblastic leukemia (12). Other BiTEs are under investigation for the treatment of solid tumors (13). Additionally, an oncolytic vaccinia virus armed with a BiTE targeting the EphA2 receptor has shown improved bystander effect and increased antitumor efficacy compared with its parental counterpart (14).
Among oncolytic viruses, adenoviruses hold features that make them good virotherapy candidates. Adenovirus replication in cancer cells can be tightly regulated, and placing transgenes under the control of the major late promoter secures replication-dependent expression (15). Furthermore, oncolytic adenoviruses have favorable toxicity profiles after systemic administration in cancer patients, and their potential is exemplified by the virus DNX-2401, which has received fast-track status and orphan drug designation by the FDA for the treatment of malignant glioma (16). We hypothesize that oncolytic adenoviruses are good candidates to deliver BiTEs to specifically redirect immune responses toward cancer cells.
In this study, we have engineered an oncolytic adenovirus to secrete an EGFR-targeting BiTE upon viral replication in cancer cells. We show that the BiTE-expressing adenovirus induces robust and specific T-cell activation and proliferation upon infection of cancer cells in vitro and in vivo. This specific T-cell activation translates into the redirected killing of cancer cells, enhancing the antitumor efficacy of the virus in mouse xenograft models of cancer. These results demonstrate that BiTE-expressing oncolytic adenoviruses have favorable properties that can overcome major limitations in oncolytic virotherapy.
Materials and Methods
Cancer cell lines A431 (epidermoid carcinoma), MDA-MB-453 (breast cancer), A549 (human lung carcinoma), HCT116 (colorectal carcinoma), FaDu (pharynx squamous cell carcinoma), Jurkat (T cell leukemia), and HEK-293 (human embryonic kidney) were obtained and authenticated by STR profiling by the ATCC. The A431-GFPLuc cell line was generated by sorting of A431 cells transduced with a lentiviral vector encoding GFP and luciferase.
Preparation of peripheral blood mononuclear and T cells
All experiments were approved by the ethics committees of the University Hospital of Bellvitge and the Blood and Tissue Bank (BST) from Catalonia. Blood samples were obtained from the BST from Catalonia and the Human Immunology Core of the University of Pennsylvania. Peripheral blood mononuclear cells (PBMC) of healthy donors were isolated by ficoll density gradient centrifugation. T cells were negatively isolated using the RosetteSep Human T Cell Enrichment Cocktail (STEMCELL Technologies). For stimulation, T cells were cultured with CD3/CD28-activating Dynabeads (Thermo Fisher Scientific) at a bead-to-cell ratio of 3. For biodistribution experiments, T cells were transduced with a lentiviral vector expressing the click beetle green luciferase [multiplicity of infection (MOI) = 5] 24 hours after activation (17). Preactivated T cells were expanded and handled as previously described (18).
cBiTE and adenoviruses
The anti-EGFR scFV (C225) was derived from the cetuximab sequence obtained from publicly available sources. The anti-CD3 scFV sequence of the Blinatumomab BiTE was obtained from patent application WO2004106381. The C225 and anti-CD3 variable regions were connected by a (G4S1)3 and a (G2S1)4GG linker, respectively, and both scFV were connected to each other by a GGGS flexible linker. The cBiTE was arranged VLC225–VHC225–VHCD3–VLCD3 and included the peptide signal from the mouse Ig heavy chain and a FLAG tag at the N- and C-terminus of the protein, respectively. The cBiTE construct was optimized for human codon usage and synthesized by Genscript (Genscript). Viruses ICOVIR-15K (abbreviated ICO15K) and AdTLRGDK have been previously described (19, 20). The cBiTE gene was incorporated into the ICO15K genome by recombineering in bacteria (21, 22). ICO15K-cBiTE was rescued after transfection of the resulting plasmid into 293 cells. Replicating viruses were propagated in A549 cells and double purified by cesium chloride gradient centrifugation. Functional [transducing units (TU)/mL] and physical [viral particles (VP)/mL] titers of purified viruses were determined by anti-hexon staining (23) and by optical absorbance at 260 nm (24), respectively.
Antibodies and flow cytometry
Flow cytometry was performed with a Gallios cytometer (Beckman Coulter). Expression of EGFR was detected with the mouse monoclonal antibody Clone 528 (Merck Millipore) or its IgG2a isotype control (Santa Cruz Biotechnology), followed by incubation with an Alexa Fluor 488–coupled goat anti-mouse IgG antibody (Thermo Fisher Scientific). For phenotyping of T-cell subpopulations, the following monoclonal antibodies coupled to different fluorochromes were used: CD3, CD4, CD8, CD69, and CD25 (Biolegend). Flow cytometry data were analyzed with the FlowJo software v7.6.5 (Tree Star).
Production of supernatants
A549 cells (1 × 107) were infected with ICO15K or ICO15K-cBiTE (MOI = 20) and supernatants were harvested 72 hours after infection. For binding assays with effector cells, supernatants were concentrated (approximately 20×) with Amicon Ultra-15 filter units with a molecular weight cutoff of 30 kDa (Merck Millipore). Supernatants from uninfected cells were used as a negative mock control.
Target (2 × 105) or effector (1 × 105) cells were incubated with the supernatants for 1 hour on ice. Cells were stained by using the monoclonal M2 anti-FLAG antibody (Sigma Aldrich) or its corresponding IgG1 isotype control (Santa Cruz Biotechnology) as primary antibodies and goat anti-mouse IgG as secondary antibody (Thermo Fisher Scientific).
In vitro coculture experiments
All coculture experiments were performed as follows, unless indicated otherwise. A total of 3 × 104 target cells/well and 1.5 × 105 PBMCs/well (E:T = 5) were seeded in 96-well plates in 100-μL medium. To assess T-cell activation by the supernatants from virus-infected cells, cocultures were mixed with 100 μL of the supernatants and incubated for the indicated times. For the oncolysis-mediated T-cell assays, cocultures were infected with ICO15K or ICO15K-cBiTE (MOI =20). For T-cell activation assays, cocultures were incubated for 48 hours or 5 days as described above and then cells were stained for cell viability with LIVE/DEAD (Thermo Fisher Scientific) followed by incubation with antibodies specific for CD8, CD4, and CD69 or CD25. For cytokine production assays, supernatants were obtained 48 hours after coculture and assessed for cytokines with the human IFNγ and TNFα ELISA (Peprotech). For proliferation assays, PBMCs were labeled with 1 μmol/L CFSE (Sigma Aldrich) and cocultured as described above for 6 or 7 days. Cells were then stained for cell viability with LIVE/DEAD followed by incubation with antibodies for CD4 and CD8.
Virus- and cell-mediated cytotoxicity assays
Virus cytotoxicity assays were performed as previously described (25). The inhibitory concentration 50 (IC50) was calculated with GraphPad Prism v6.02 (GraphPad Software Inc.) by a dose–response nonlinear regression with a variable slope.
CFSE-labeled target cells (A431 or MDA-MB-453, 1.5 × 104) were cocultured with 1.5 × 105 PBMCs (E:T = 10) in 96-well plates. Cocultures were mixed with the supernatants (100 μL) and incubated for 24 hours. Cocultures were trypsinized and stained with 10 μg/mL 7-amino-actinomycin D (7-AAD; Enzo Life Sciences). Cells were analyzed by flow cytometry, and the percentage of CFSE+/7-AAD+ cells was determined.
For bystander killing assays, A549 cells in suspension were infected with ICO15K or ICO15K-cBiTE (MOI = 20) for 4 hours. Excess virus was washed with PBS. 2.5 × 104 virus-infected A549 cells per well, and 2.5 × 104 A431-GFPLuc per well were cocultured with 1.5 × 105 PBMCs (E:T = 6 with respect to A431-GFPLuc cells) in 96-well plates. Cocultures without PBMCs were used as a control for virus-mediated cell cytotoxicity. Five days after infection, cocultures were stained with 7-AAD as described above. Cells were analyzed by flow cytometry and the absolute number of live A431-GFPLuc was determined using CountBright Absolute Counting Beads (Thermo Fisher Scientific). The percentage of live A431-GFPLuc was obtained by normalizing the absolute count of the samples to a culture of untreated A431-GFPLuc cells.
Oncolysis-mediated enhanced cell-mediated cytotoxicity was assessed with the xCELLigence Real-Time Cell Analyzer (ACEA Biosciences). A total of 1 × 104 target cells per well (A549 or HCT116) were seeded and infected with ICO15K or ICO15K-cBiTE at an MOI of 1. After overnight cell adherence, effector cells (preactivated T cells) were added (E:T = 5). Cell index (i.e., relative cell impedance) values were monitored every 20 minutes for 120 hours and normalized to the maximal cell index value immediately prior to effector cell plating.
In vivo studies
All animal experiments were approved by the Ethics Committee for Animal Experimentation from the Biomedical Research Institute of Bellvitge (IDIBELL). Animals were housed in the IDIBELL Animal Core Facility (AAALAC Unit 1155). Subcutaneous A549 and HCT116 were established by injecting 5 × 106 and 2.5 × 106 cells, respectively, into both flanks of 8-week-old female SCID/beige mice (Envigo). When tumors reached ∼100 mm3, mice were randomized and treated as described below.
For the T-cell biodistribution study, tumors were injected with 2 × 109 VP of the indicated viruses or PBS. After 5 days, mice received an intravenous injection of 1 × 107 cell preparations containing 59% preactivated CBG luciferase-expressing T cells, followed by an intraperitoneal dose of 1500 IU IL2 (Peprotech). Mice were imaged daily until day 9 with the IVIS Lumina XRMS Imaging System (PerkinElmer) after administering intraperitoneal injections of a 15 mg/mL d-luciferin firefly potassium salt solution (Biosynth AG). Tumor radiance was measured by drawing a region of interest around the tumor contour.
For antitumor efficacy, A549 tumors were injected with 2 × 109 VP of the indicated viruses or PBS. After 4 and 18 days, PBS or 1 × 107 unstimulated human PBMCs were administered to the mice by intravenous injection. Mice bearing HCT116 tumors received an intravenous injection of 1 × 1010 VP of the indicated viruses or PBS. After 4, 8, and 11 days, mice received an intravenous injection of 1 × 107 preactivated T cells followed by an intraperitoneal injection of 1,500 IU IL2. In both models, tumors were measured every 2 to 4 days, and the volume was calculated as V (mm3) = π/6 × W2 × L, where W and L are the width and the length of the tumor, respectively. Immunohistofluorescence of OCT-embedded sections of A549 tumors at the end of the study was used to evaluate the expression of the E1a protein as described (26).
mRNA extraction and real-time PCR
Frozen tumor samples were disrupted using a mortar and pestle under liquid nitrogen. Approximately 25 mg of tissue was homogenized with Qiashredder homogenizers, and RNA was isolated with the RNeasy kit (Qiagen), with DNase I digestion to remove genomic DNA. RNA (1 μg) was retrotranscribed with the High-Capacity cDNA Reverse Trancription kit (Thermo Fisher Scientific).
Real-time analysis was performed using a LightCycler 480 Instrument II (Roche) in the presence of SYBR Green I Master (Roche). PCR conditions were: 95°C 10 minutes, 40× cycles of 95°C 15 seconds, 60°C 1 minute and 72°C 7 seconds. Hexon primers were Ad18852 5′-CTTCGATGATGCCGCAGTG-3 and Ad19047R 5′-ATGAACCGCAGCGTCAAACG-3′. cBiTE primers were qBiTEF 5′-CGGCGAGAAAGTGACAATGAC-3′ and qBiTER 5′-TTGGTGAGGTGCCACTTTTC-3′. Standard curves for cBiTE and hexon were prepared by serial dilutions of known copy numbers of pUC57-cBiTE or purified ICO15K genomes, respectively. Non-retrotranscribed RNA samples, equivalent to the amount cDNA loaded in the PCR, were run to discard genomic DNA contamination.
For comparisons of two groups, two-tailed unpaired t tests were used. For comparison of more than two groups, one-way ANOVA with Tukey post hoc tests was used. Statistical significance was established as P < 0.05. Data are presented as the mean ± SD or SEM. All statistical analyses were calculated with the GraphPad Prism software v6.02.
Generation and characterization of a BiTE-armed oncolytic adenovirus targeting the EGFR
We have reported the generation of ICOVIR-15K (ICO15K), an E1a-Δ24–based oncolytic adenovirus with palindromic E2F binding sites in the E1a promoter and an RGDK motif replacing the KKTK heparan sulfate glycosaminoglycan-binding domain in the fiber shaft (19). This virus has shown favorable toxicity profiles and increased tumor targeting in vivo. We engineered ICO15K to express an EGFR-targeting bispecific T-cell engager (cBiTE) under the control of the adenovirus major later promoter (Fig. 1A). This location was chosen to avoid potential BiTE-mediated interference with virus replication. The virus ICO15K-cBiTE was successfully rescued and it retained oncolytic properties in vitro, despite a decrease in the IC50 values compared with its parental counterpart in dose–response cytotoxicity assays (Fig. 1B).
To evaluate the secretion of the cBiTE by infected cells, flow cytometry–based binding assays were performed by exploiting the FLAG tag encoded in the transgene. For these assays, a panel of cancer cell lines with varying EGFR expression levels, and CD3+ Jurkat and human PBMCs were used (Supplementary Fig. S1). cBiTE antibodies were detected only in the supernatants of ICO15K-cBiTE–infected cells and they specifically bound to EGFR+ (A431, A549, HCT116, and FaDu) but not EGFR− (MDA-MB-453) cancer cells (Fig. 2A and Supplementary Fig. S2). cBiTE molecules also bound to CD3+ Jurkat cells and human PBMCs, and this binding was more pronounced when supernatants were concentrated approximately 20× (Fig. 2B). Furthermore, the binding of the cBiTE to CD4+ and CD8+ T cells within PBMCs was confirmed (Fig. 2C).
Supernatants from ICO15K-cBiTE–infected cells enhance T-cell function
We then evaluated the functionality of the secreted cBiTE in coculture experiments of cancer cells with unstimulated human PBMCs. ICO15K-cBiTE supernatants specifically induced CD8+ and CD4+ T-cell activation, as indicated by an increase in the expression of the activation markers CD25 and CD69 when PBMCs were cocultured with EGFR+ cell lines (Fig. 3A). To confirm this activation, we assessed cytokine production by PBMCs in coculture assays with a panel of cancer cell lines. IFNγ and TNFα were detected at high levels only when ICO15K-cBiTE supernatants were cocultured with PBMCs and EGFR+ cell lines (Fig. 3B). Importantly, the cBiTE-containing supernatants did not induce T-cell activation when cocultured alone or with the EGFR-negative cell line MDA-MB-453.
Another important indicator of T-cell activation is their proliferative capacity. CD3+ T cells underwent multiple rounds of proliferation only when cocultured with EGFR+ A431 cells and ICO15K-cBiTE supernatants, as evidenced by the dilution of CFSE after 5 days of incubation (Fig. 3C).
The ultimate goal of BiTE antibodies is to retarget T-cell–mediated cytotoxicity toward cancer cells. To confirm this, we performed cell-mediated cytotoxicity assays by coculturing A431 and MDA-MB-453 CFSE-labeled cells with PBMCs and the different supernatants for 24 hours. The coculture of A431 but not of MDA-MB-453 cells with the cBiTE-containing supernatants and PBMCs led to a significant increase in cell cytotoxicity compared with ICO15K and the mock control (Fig. 3D).
ICO15K-cBiTE–mediated oncolysis enhances T-cell function and induces a T-cell–mediated bystander effect
The experiments described above were performed with supernatants of infected cells that contained the cBiTE. We next wanted to evaluate the cBiTE-expressing virus in a setting that more closely resembles the oncolytic process. For this, cocultures of human PBMCs with either A549 or HCT116 EGFR+ cell lines were infected at an MOI of 20. Cells were incubated for 5 and 7 days to address activation and proliferation, respectively. These incubation times were chosen to allow a full replication cycle of the virus and the proper action of the cBiTE. ICO15K-cBiTE–mediated oncolysis led to an increase in CD25+ and CD69+ T cells and to extensive proliferation of both T-cell subsets only when cancer cells were present (Fig. 4A and B).
Another important feature of a secreted BiTE is its bystander effect. To test this without the interfering cytotoxicity of the virus, we chose the EGFR+ A431-GFPLuc cell line, which expresses low levels of the coxsackie- and adenovirus receptor. This cell line is partly resistant to adenovirus infection and shows low adenovirus-mediated cytotoxicity and no cBiTE production at high MOIs (Supplementary Fig. S3). A431-GFPLuc cells were cocultured with ICO15K- or ICO15K-cBiTE–infected A549 cells in the presence or absence of human PBMCs for 5 days. Thus, in this setting, A549 cells act as BiTE producers, while A431-GFPLuc cells represent the targets of the T cells. ICO15K-cBiTE–infected A549 cells induced a significant decrease in the percentage of live A431-GFPLuc cells compared with ICO15K-infected cells when cocultured with PBMCs from two different donors (Fig. 4C). This increased cytotoxicity was dependent on the presence of PBMCs, as ICO15K and ICO15K-cBiTE induced similar levels of cell death in the absence of PBMCs.
We then tested whether the infection of cancer cells with ICO15K-cBiTE at low MOIs could provide an advantage during oncolysis over the nonmodified virus when cocultured with preactivated T cells. Preactivated T cells have a phenotype that more closely resembles that of adenovirus-specific CTLs, the main T-cell population expected to infiltrate tumors during oncolysis. We found that preactivated T cells show enhanced cytotoxic potential compared to naïve T cells (Supplementary Fig. S4A) and produce high amounts of proinflammatory cytokines (Supplementary Fig. S4B) when engaged by cBiTE antibodies. T cells were added to the cultures of HCT116- or A549-infected cells and cell death was assessed using the xCELLigence system. The combination T cells and ICO15K-cBiTE showed a remarkable additive effect by reducing by half the time to complete death of A549 and HCT116 cells when compared with any of the other virus treatments (Fig. 4D).
ICO15K-cBiTE increases the persistence and accumulation of tumor-infiltrating T cells
To address the potential of ICO15K-cBiTE in vivo, we first evaluated the persistence and biodistribution of human T cells in SCID/beige mice. To this end, human T cells were preactivated and transduced with a lentiviral vector encoding the Click Beetle Green (CBG) luciferase and GFP. T-cell preparations had 59% transduced (i.e., GFP+) cells and retained the ability to proliferate in vitro (Supplementary Fig. S5). To test the biodistribution and persistence of these Luc-T cells in vivo, SCID/beige mice bearing subcutaneous HCT116 tumors were injected intratumorally with PBS, ICO15K-cBiTE, or the parental virus. Five days after virus administration, mice received an intravenous injection of the CBG-Luc T-cell preparations and tumor luminescence was measured daily by in vivo imaging. T cells were mainly distributed in the lymphoid organs 1 day after infusion and the signal generally decreased within 3 days. Tumor radiance was measured from day 3 on to avoid background signal coming from the initial engraftment in the lymphoid organs. Tumors treated with ICO15K-cBiTE–induced T-cell accumulation from day 3 with a peak at day 4, and with the signal lasting for 9 days until reaching the levels of PBS- or ICO15K-treated tumors (Fig. 5A). Half of the tumors treated with ICO15K-cBiTE, but none of those treated with PBS or ICO15K, showed a detectable increase in bioluminescent signal after 3 days of adoptive T-cell transfer (Fig. 5B). In a similar independent experiment, virus and cBiTE transcripts were detected in tumors upon intratumoral administration of virus and subsequent systemic administration of T cells (Fig. 5C).
ICO15K-cBiTE enhances antitumor efficacy in vivo
We then assessed the antitumor efficacy of ICO15K-cBiTE in mouse xenograft models of cancer. SCID/beige mice bearing subcutaneous A549 tumors were injected intratumorally with PBS, ICO15K-cBiTE, or the parental virus. At 4 and 18 days after virus administration, mice received intravenous injections of either unstimulated human PBMCs or PBS. The PBS groups showed the fastest tumor growth and the cBiTE-expressing adenovirus showed a similar antitumor efficacy as the parental virus in the absence of PBMCs. Notably, the administration of human PBMCs significantly enhanced the antitumor efficacy of ICO15K-cBiTE but not that of ICO15K (Fig. 6A). Histologic analysis of the tumors revealed the expression of the E1a protein at day 49 in all groups treated with any of the adenoviruses in the study, indicating that PBMC-mediated cytotoxicity had no effect on the persistence of the virus in the tumor (Fig. 6B).
We also tested the effect of delivering ICO15K-cBiTE intravenously, a setting that is more therapeutically relevant. SCID/beige mice bearing subcutaneous HCT116 tumors were injected intravenously with PBS, ICO15K, or ICO15K-cBiTE. On days 4, 8, and 11 after virus administration, mice received an intravenous injection of preactivated T cells followed by an intraperitoneal injection of IL2. This regime was chosen based on the peak of persistence at day 4 (Fig. 5A), and it aimed at maximizing the presence of T cells in the tumor during the first weeks after virus treatment. Adenovirus-treated mice had a smaller tumor size than those treated with PBS, with ICO15K-cBiTE–treated ones having the smallest tumor volume after T-cell administration (Fig. 6C). Although these differences were not statistically significant for tumor volume, ICO15K-cBiTE showed significant reduction in tumor growth compared with PBS and ICO15K from day 7 after virus administration (Fig. 6D). The presence of virus and cBiTE in tumors was confirmed at the end of the study by immunohistologic analysis (Supplementary Fig. S6).
In this study, we armed the oncolytic adenovirus ICO15K with an EGFR-targeting BiTE (cBiTE). cBiTEs secreted from infected cells retained key features of BiTEs including target cell–dependent T-cell activation and proliferation and redirected lysis of cancer cells (27, 28). The anti-EGFR scFV in the cBiTE was derived from the monoclonal antibody cetuximab, which is used in patients with colorectal and head-and-neck squamous cell cancer (29, 30). One of the major mechanisms of resistance to cetuximab in colorectal cancer is the mutation of downstream signaling genes such as BRAF, KRAS, PIK3CA, and PTEN (31). A cetuximab-derived BiTE overcomes this resistance by successfully redirecting T cells to kill BRAF- and KRAS-mutated colorectal cancer cells (32). Here, we also demonstrate that ICO15K-cBiTE induces a T-cell–mediated killing of KRAS-mutated HCT116 cells in vitro and in vivo.
The potential of BiTE-armed oncolytic viruses has been previously demonstrated using an oncolytic vaccinia virus with an EphA2-targeting BiTE (vv-EphA2; ref. 14). Whereas vv-EphA2 induced T-cell activation and PBMCs-mediated bystander killing of cancer cells, no T-cell proliferation was observed in the absence of exogenous IL2 in vitro and in vivo. In contrast, we observed that ICO15K-cBiTE induced remarkable proliferation of T cells in vitro, without the addition of IL2 to the cocultures. This difference could be related to the BiTE design or the oncolytic virus used. Another noteworthy finding from our in vivo experiments was the increased T-cell infiltration and expansion observed in subcutaneous tumors treated with the cBiTE-expressing virus after intravenous T-cell administration. With regard to antitumor efficacy, vv-EphA2 prevented tumor growth of subcutaneous A549 tumors that were coimplanted with PBMCs immediately followed by an intraperitoneal injection of the virus, and delayed growth of 7-day established lung A549 tumors after the coadministration of vv-EphA2 and PBMCS. In our antitumor efficacy experiments, we chose to use tumors established for longer periods (three and two weeks for A549 and HCT116, respectively), and efficacy was observed even in this setting. Altogether, our work confirms the potential of arming oncolytic viruses with BiTEs. Leaving aside the different tumor targets used in these approaches (EphA2 or EGFR), vaccinia could offer a faster and more immunogenic oncolysis, whereas adenovirus a higher selectivity based on pRB pathway constitutively activated in tumor cells.
The strategy of arming oncolytic adenoviruses with BiTEs exploits the best features of both therapies simultaneously, while overcoming many of the hurdles encountered by both as single agents. From the point of view of the BiTE, its localized expression from infected cancer cells would reduce the adverse effects commonly observed with BiTEs, some of which can be fatal if not treated appropriately (33). Additionally, the continuous production of BiTEs from infected cancer cells would increase its availability at the tumor site. This would be advantageous given the short half-life of BiTEs in the serum, which requires continuous drug infusion (i.e., 4–6 weeks; ref. 34). Conversely, BiTEs could help to overcome some of the limitations in oncolytic virotherapy. Although oncolytic adenoviruses have been shown to elicit specific T-cell responses to tumor neoepitopes in vivo (35), the immunodominance of adenoviral epitopes masks immune responses to delivered transgenes or capsid-displayed tumor antigens (36, 37). We and others have successfully developed approaches to favor MHC-I–restricted antitumor rather than antiviral immune responses (7, 38). However, these strategies rely on MHC-I expression, which is downregulated on cancer cells as one mechanism to escape from immune responses (4, 39). The MHC-I–independent and polyclonal mode of action of BiTEs (40) offers a unique opportunity to theoretically activate and redirect any tumor infiltrating T cell, including adenovirus-specific CTLs, to cancer cells. Interestingly, it has been recently reported that BiTE antibodies can engage cytomegalovirus-specific CTLs to kill cancers cells in vitro (41). Because we have demonstrated the cBiTE-mediated polyclonal engagement of naïve and preactivated CD4+ and CD8+ T cells toward cancer cells, we speculate that redirecting adenovirus-specific CTLs is also feasible. In addition to the antiviral immune responses, limited virus spread within the tumor due to stromal barriers is another limitation for oncolytic adenoviruses (1, 42). The small size of BiTEs (∼55 kDa) is advantageous for penetration and distribution throughout the tumor as demonstrated by imaging studies (43). This would allow improving the overall bystander effect of oncolytic adenoviruses even if their spread is limited. Furthermore, this strategy can be expanded to attack the tumor stroma by using BiTEs targeting, for example, the fibroblast activation protein alpha on cancer-associated fibroblasts. Such a BiTE has been validated in vitro (44), and its expression from infected cells would focus the bystander effect to stromal cells, which are normally resistant to the oncolytic effect of the virus.
Although we were able to amplify ICO15K-cBiTE, its onocolytic properties were reduced approximately 2-fold compared with the nonmodified virus. This loss in cytotoxicity may be the result of the competition between the cBiTE and viral genes for transcription and translation. Despite this, the cBiTE gene gives a remarkable advantage to the virus in the presence of T cells even at low MOI in vitro. Importantly, the loss in in vitro cytotoxicity of the cBiTE-expressing virus did not translate into a loss in antitumor efficacy in vivo in the absence of PBMCs.
We evaluated ICO15K-cBiTE systemically, which is the preferred route of administration for oncolytic adenoviruses to maximize the delivery of the virus to metastatic tumors. Furthermore, we have described that replacing the KKTK domain in the shaft of the fiber with an RGD motif enhances the systemic antitumor efficacy of ICO15K by reducing liver targeting and increasing virus bioavailability in the blood (19). In our study, the systemic administration of ICO15K to SCID/beige mice led to a modest decrease in tumor growth compared with PBS. This modest antitumor efficacy is most likely explained by the low dose of virus administered. It has been demonstrated that SCID/beige mice are more sensitive to oncolytic adenovirus infection than BALB/c nude mice due to their increased immunodeficiency (45). Based on our preliminary studies, we determined that the highest systemic tolerated dose of our viruses in SCID/beige mice was 1 × 1010 VP/mouse (data not shown). This dose is five times lower than the one used for efficacy studies of ICO15K in nude mice, in which a clear advantage was observed in several mouse models of cancer. Note, however, that this low dose was enough to observe the advantage of ICO15K-cBiTE in the tumor model we evaluated.
Because the EGFR is expressed on cancer cells targeted by the virus, we were concerned about the potential interference of the cBiTE-mediated cell killing with virus replication and persistence in vivo. E1a expression was detected in all virus-treated tumors at the end of the study and it was independent of PBMC administration. These results show that, at least in our model, the cBiTE-mediated cancer cell death does not compromise the persistence of the virus in the tumor. It is worth mentioning the limitation of our model due to the transient persistence of human T cells in SCID/beige mice. An ideal setting to prove the persistence of the virus in the tumor would involve the use of immunocompetent mouse models. Furthermore, such a model would enable to study the ability of virus-delivered BiTEs to activate infiltrating T cells under the highly immunosuppressive environment of the tumor. However, the lack of adenovirus replication in murine cells limits the study of adenovirus-mediated oncolysis in such models. Despite the cumbersome models of PBMC transfer in SCID/Beige mice with human tumor xenografts, it is worth highlighting that a BiTE-armed oncolytic adenovirus represents a single-agent systemic treatment easy to translate in patients where the immune response against the virus provides a continuous source of endogenous T cells to the tumors, enhancing the potential efficacy.
The marketing approval in the United States of the oncolytic virus T-VEC for the treatment of advanced melanoma encourages the development of oncolytic viruses with improved immunotherapeutic potential. Our data demonstrate that BiTE-armed oncolytic adenoviruses hold the unique properties of inducing specific and redirected antitumor immune responses. This strategy has the potential to solve key limitations in oncolytic virotherapy and encourages its further evaluation and development.
Disclosure of Potential Conflicts of Interest
C.H. June reports receiving a commercial research grant from Novartis. No potential conflicts of interest were disclosed by the other authors.
Conception and design: C.A. Fajardo, S. Guedan, R. Alemany
Development of methodology: C.A. Fajardo, S. Guedan, R. Moreno, R. Alemany
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.A. Fajardo, L.A. Rojas, M. Arias-Badia
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.A. Fajardo, L.A. Rojas, C.H. June, R. Alemany
Writing, review, and/or revision of the manuscript: C.A. Fajardo, S. Guedan, C.H. June, R. Alemany
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Arias-Badia
Study supervision: R. Alemany
Other (helped carrying out some of the experiments): J. de Sostoa
The authors thank Brian Keith for helpful comments on the article.
This work was supported by a BIO2014-57716-C2-1-R grant from the Ministerio de Economía y Competitividad of Spain and a 2014SGR364 research grant from the Generalitat de Catalunya. C.A. Fajardo was supported by the European Commission Marie Curie Initial Training Network ADenoViruses as novel clinical treatments (ADVance, FP7; project reference: 290002). Co-funded by the European Regional Development Fund, a way to Build Europe. S. Guedan and C.H. June are members of the Parker Institute for Cancer Immunotherapy, which supported the University of Pennsylvania Cancer Immunotherapy Program.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.